Bankruptcy prediction gives an overview to management and investors about actual condition of the company briefly and clearly, because the poor condition does not happenall of a sudden it can be seen regularly. Bankruptcy prediction in this study uses three approaches, namely the model Z-score, Springate, and Zmijewski. This study is aimed to determine the metals and manufacturing companies’ health level which listed in Indonesia Stock Exchange and also find out if there are significant differences of the three models were used to examine the approach of a company defaults. This study uses secondary data of 11 companies in th ebasic industry and the metal type and etc. Linear regression and T-test are used to prove the hypothesis. The results of this study concludes that there is no significant difference to the prediction of model Z-Score and Springate, but there is a significant difference to the prediction model Z-Score with Zmijewski and Springate with Zmijewski. Z-Score Model predictions show many companies are in a state of potential bankruptcy, as well as Springate, but Zmijewski stating that the company more healthy.